Biochem. J. (1984) 224, 829-837Printed in Great Britain

Cyanide inhibition of cytochrome c oxidase

A rapid-freeze e.p.r. investigation

Peter JENSEN, Michael T. WILSON,* Roland AASA and Bo G. MALMSTROMDepartment of Biochemistry and Biophysics, University of Gothenburg and Chalmers University of

Technology, S-412 96 Gioteborg, Sweden

(Received 14 May 1984/Accepted 29 August 1984)

The inhibition of cytochrome c oxidase by cyanide, starting either with the resting orthe pulsed enzyme, was studied by rapid-freeze quenching followed by quantitativee.p.r. It is found that a partial reduction of cytochrome oxidase by transfer of 2 elec-tron equivalents from ferrocytochrome c to cytochrome a and CuA will induce a tran-sition from a closed to an open enzyme conformation, rendering the cytochrome a3-CUB site accessible for cyanide binding, possibly as a bridging ligand. A heterogeneityin the enzyme is observed in that an e.p.r. signal from the cytochrome a33+-HCNcomplex is only found in 20% of the molecules, whereas the remaining cyanide-bounda3-CuB sites are e.p.r.-silent.

The inhibition of cytochrome c oxidase (EC1.9.3.1) by cyanide remains an interesting problemhalf a century after its discovery. The presentphase of study of this phenomenon was initiatedby Van Buuren (1972), who in his comprehensivethesis summarized the earlier findings and also re-investigated the action ofcyanide on the oxidase inorder to establish the site of inhibition and itsredox state. Among the four redox centres in theenzyme (cytochrome a and a3, CUA and CUB), cyto-chrome a3 has long been considered the bindingsite for cyanide as well as for many otherinhibitors. A dilemma, however, has been theobservation that the rate of cyanide binding toeither the fully oxidized or the fully reducedenzyme is too slow to account for the rate of onsetof inhibition in turnover experiments or withsystems in vivo; recent work (Jones et al., 1984) hasestablished that the rates differ by as much as afactor larger than 105. From his experiments, VanBuuren (1972) concluded that the species suscept-ible to rapid inhibition is the partially reducedoxidase. He also suggested that partial reductioncauses a conformational change that makes oxid-ized cytochrome a3 more available to cyanide.

In stopped-flow experiments with reduced cyto-chrome c and 02 (Antonini et al., 1977) it was

* Visiting scientist of the Nobel Institute for Chemis-try, 1983. Permanent address: Department of Chemis-try, University of Essex, Wivenhoe Park, Colchester,Essex C04 3SQ, U.K.

found that cytochrome oxidase can appear in twoforms, which are optically similar but kineticallydifferent: the 'resting' and the (kinetically faster)'pulsed' form. The term 'pulsed' refers to theenzyme that has been freshly oxidized by thereaction of the reduced enzyme with 02- In theabsence of reducing substrates it slowly decaysback to the resting form (Brunori et al., 1981). Thepulsed oxidase is related to another highly activeenzyme form found by Sekuzu et al. (1959) andnamed 'oxygenated' oxidase by them. Thesefindings opened the possibility that the enzymeforms with high activity are also the conforma-tional states that bind cyanide rapidly, as had infact been suggested by Brittain & Greenwood(1976) for the 'oxygenated' species.

This question has recently been addressed indetail in a re-investigation by Jones et al. (1984) ofthe kinetics of cyanide binding. They showed thata model with only two conformational states,resting and pulsed, is inadequate. An extendedmodel was suggested, in which the enzyme in thepulsed as well as in the resting state can exist in a'closed' or an 'open' conformation, of which onlythe latter reacts rapidly with cyanide. It wasfurther proposed that the transition from the closedto the open conformation is triggered by thereduction of cytochrome a and CuA (and perhapsalso CUB), and that the conformational change isfollowed by a rapid binding of cyanide to oxidizedcytochrome a3.

Although the model proposed by Jones et al.

Vol. 224

829

P. Jensen, M. T. Wilson, R. Aasa and B. G. Malmstrom

(1984) can adequately account for their extensivekinetic results, it would seem desirable to havemore direct physical evidence to test the validity ofits details. To this end we have followed thereaction of reduced cytochrome c with resting andwith pulsed oxidase in the absence and in thepresence of cyanide by e.p.r. spectroscopy with theaid of a rapid-freeze quenching technique. Ourdata show that the majority of the enzymemolecules in the resting oxidase react rapidly withcyanide when cytochrome a and CUA have beenreduced, and this prevents entry of more than 2electron equivalents from reduced cytochrome c.Also, with the pulsed oxidase only 2 electronequivalents can enter the enzyme in the presenceof cyanide. In agreement with observations byBrudvig et al. (1981), we find, however, that theresting oxidase is heterogeneous. About 20% ofthe molecules are in a form in which the cyanide-inhibited enzyme has accepted 3 electron equi-valents, as evidenced from the appearance of a low-spin Fe3+ e.p.r. signal from cytochrome a3-HCN,showing that CUB must be reduced; this fraction ofthe molecules gives a high-spin Fe3+ signal in theabsence of cyanide. Our results also indicate thatthe pulsed oxidase is heterogeneous, some mole-cules reacting rapidly with cyanide before bothcytochrome a and CUA have been re-oxidized.

Materials and methodsEnzyme and chemicalsCytochrome c oxidase was prepared from ox

heart by the method of Van Buuren (1972). Theenzyme contained 10nmol of haem A and0.1-0.15mg of phospholipids (measured as lipid-bound P)/mg of protein. Absorbance ratios(typical) were: A445(red.)/A42O(red.) > 2.3, A605(red.)/A554(red.)> 2.3, A445(red.)/A605(red.)> 5.0. The mol-ecular activity was 20s-1, measured essentially asdescribed by Smith & Conrad (1956). When thepurified enzyme was incorporated into asolectinvesicles the activity rose to around 400s-1 (pH 6.2,high ionic strength).Enzyme concentrations were determined as

described in Jensen et al. (1981). Sodium dodecylsulphate/polyacrylamide-gel electrophoresis by theprocedure of Downer et al. (1976) gave a subunitcomposition similar to that described by Wikstromet al. (1981) (seven major subunit bands).

Horse heart cytochrome c (type III) from SigmaChemical Co. was used without further purifica-tion. Stock solutions of 2mM and 2.5mM reducedcytochrome c were prepared by titration withsodium ascorbate until the absorption band ofcytochrome c at 695nm was completely bleached.The solution was then equilibrated with air or pure02. Inspection of the 695nm band showed no signs

of re-oxidation. The amount of polymerized cyto-chrome c was less than 2%, determined from thedifference in absorbance between ascorbate anddithionite-reduced cytochrome c at 550nm. Trans-ferrin was kindly given by Dr. P. Aisen.

All experiments were performed in 50mM-Hepes[4-(2-hydroxyethyl)- 1 -piperazine-ethanesulphonicacid]/NaOH buffer, pH 7.4, at room temperature.Stock solutions of 0.48M-KCN were titrated withHCl to neutrality before use.

All chemicals were of analytical grade. Allsolutions were prepared fresh on the day of theexperiment.

SpectroscopyOptical spectra were recorded in a Beckman

Acta MIV spectrophotometer. E.p.r. spectra at15K and 9GHz were recorded with a Varian E-9spectrometer equipped as described in Wilson etal. (1982). Four to eight spectra were collected fromeach sample and time-averaged to improve thesignal-to-noise ratio before integration.

Rapid-freeze experimentsThe rapid-freeze equipment and freezing proce-

dures were identical with the description given inWilson et al. (1982) except that the optical cell wasused only as a container for the enzyme solutionafter mixing when delays were used. The T-mixerwas also replaced by a tangential rapid-flow mixer,similar to the one described by Ballou & Palmer(1974). Gas-tight 2.5 ml Hamilton syringes wereused throughout the experiments (equal mixing).The experiments starting with the resting en-

zyme were performed as follows. Syringe A wasfilled with 246pM-cytochrome oxidase solution inbuffer equilibrated with air, and syringe B wasfilled with buffer only. An oxidized referencesample was then collected by mixing the contentsof the two syringes in equal proportions. Thebuffer in syringe B was then replaced with 2mMaerated reduced cytochrome c. After collection ofone sample without delay the cell compartmentwas attached and samples with different delaytimes were collected by using the multiple-pushfacility of the system. KCN was then injected insyringe B from the concentrated stock solution togive a final concentration of 20mM. The samenumber of samples were collected with the samedelays as above.

Experiments with the pulsed enzyme wereperformed by reducing the cytochrome oxidase insyringe A (new enzyme preparation; 345 Mm) withan equivalent amount of sodium ascorbate andwith a catalytic amount of cytochrome c present.After a reference sample had been collected bymixing cytochrome oxidase with the aeratedbuffer, syringe B was filled with 02-bubbled

1984

830

Cyanide inhibition of cytochrome c oxidase

2.5mM reduced cytochrome c solution, andsamples were taken as above. The cyanide additionwas also made as with the resting enzyme. Asinternal standard 100 uM-transferrin (see below)was added to the cytochrome oxidase solution insyringe A before the pulsed experiments.A second type of experiment with the pulsed

enzyme was carried out by just mixing the reducedcytochrome oxidase with aerated cyanide-contain-ing buffer with the same concentrations and delaysas above.

Quantitative measurement of the rapid-freezesamples

It is for many reasons inevitable that difficultiesarise when making quantitative e.p.r. measure-ments of rapid-freeze samples. The main problemsoccur during the packing of the powdered samples,where the sample concentration after packingcan vary. This difference is due to variations inthe packing force applied on the sample, andits packing characteristics. From studies ofCuSO4/EDTA solutions Ballou (1971) found avariation in packing factor of 6% with the sameperson packing the samples (packing factor de-fined as the ratio between the signal from a slowlyfrozen sample and a rapid-freeze one of the samematerial). It should be noted that this study wascarried out with aqueous solutions of small ions,which pack easily. Uniform protein solutions posesno problems in packing, but, if chemical reactionsthat change the shape ofthe protein can take place,it can have large effects on packing ability. Duringthe course of the experiment it was noted thatsamples captured during turnover were moredifficult to pack compared with the restingoxidized cytochrome oxidase. Thus introducingdelays before the reaction mixture was freeze-quenched led to such packing difficulties that wewere forced to decrease the jet speed by 30% to beable to pack the samples at all. An interestingobservation was made in that these difficultieswere less pronounced after cyanide addition,possibly indicating an altered protein conforma-tion after cyanide binding to the turnover enzyme.

Isopentane adhering to the outside of the e.p.r.tubes after freeze-quenching also causes varia-tions, as it can be difficult to wipe off withoutheating the sample by the resultant friction.Remaining isopentane can lead to increased frostattraction on the tubes from humid air, thuslowering the sensitivity of the e.p.r. measurement.To overcome these difficulties a small amount of

transferrin was added to the pulsed cytochromeoxidase as an internal standard. Transferrin is aniron-containing protein (Aasa et al., 1963) display-ing a strong e.p.r. signal at g = 4.2 and a weakerone at g = 9, lines that are well spaced and do not

interfere too much with the cytochrome oxidasee.p.r. signals. However, when cyanide was addedto the cytochrome oxidase solution, the transferrinsignal at g = 4.2 overlayed the g = 3.55 cyanidesignal, which made quantification difficult. Fortu-nately, though, the transferrin signal could besubtracted digitally without disturbing the rest ofthe e.p.r. spectrum. The amplitude of the g = 4.2signal of transferrin added to the samples wasmeasured, and this value was used as an internalstandard to correct the ensuing integrations of thee.p.r. signals of the samples for small variations insample temperature, instrumental sensitivity andsmall differences in packing between samples.These integrations were made by the procedure ofAasa & Vanngard (1975), except for the cyto-chrome a signal, for which instead the amplitudeof the gx = 1.45 signal was measured relative to theoxidized control sample. The g values for the low-spin a3-HCN signal was calculated by the methodof De Vries & Albracht (1979).

Results and discussionRapid-freeze experiments

Figs. 1 and 2 show e.p.r. spectra of samplesobtained from experiments in which cytochrome coxidase, as either the resting (Fig. 1) or the pulsed(Fig. 2) form, was mixed with its substrates, ferro-cytochrome c and 2, in the presence or in theabsence of KCN, and then quenched by the rapid-freeze technique. As a comparison and for quanti-tative analysis each Figure shows a controlexperiment in which the enzyme, again in eitherthe resting or the pulsed form, was quenched in theabsence of cytochrome c. These Figures also showspectra of samples captured either directly follow-ing mixing (no delay) or 5s after mixing.The signals seen in these spectra are typical of

the enzyme and have been reported elsewhere(Aasa et al., 1976; Johnson et al., 1981). Thus theg = 3 (low-spin ferric cytochrome a) and g = 2(cupric CUA) signals are seen in the controls and, tovarious extents, in the enzyme after mixing withsubstrates. In the experimental samples ferricyto-chrome c, produced as a consequence of electrontransfer to the enzyme, makes the dominantcontribution to the g = 3 region. In addition, theg =6 signal of high-spin ferric cytochrome a3appears in some of the spectra of the partiallyreduced enzyme.One of the important results of the experiments

may be readily appreciated by a qualitativeinspection of Figs. 1 and 2. It is seen that for bothresting and pulsed enzyme the addition of cyanidemakes very little difference to the e.p.r. signalsdisplayed by samples frozen at very short timesafter mixing, i.e. the general features of spectra B

Vol. 224

831

P. Jensen, M. T. Wilson, R. Aasa and B. G. Malmstr6m

g-value12 8 6 5 4

0.1

3 2

0.2 0.3Magnetic flux density (T)

1.5

0.4 0.5

Fig. 1. E.p.r. spectra from rapid-freeze experiments with resting oxidized cytochrome oxidase versus reduced cyto-chrome c

Spectrum A: resting control sample, 123 pM-cytochrome oxidase (after mixing with Hepes buffer). Spectrum B:cytochrome oxidase mixed with cytochrome C2+; no delay after mixing; cytochrome oxidase concentration, 123 gm;cytochrome c concentration, 1 mm. Spectrum C: as for spectrum B but with lOmM-KCN (added in the cytochrome csyringe). Spectrum D: as for spectrum B but with 5 s delay after mixing. Spectrum E: as for spectrum C but with 5 sdelay after mixing. All concentrations are after mixing. The arrow indicates the signal due to low-spin ferricyto-chrome a3-cyanide complex at g = 3.55. The magnified signals at g,, 1.45 is due to cytochrome a. E.p.r. conditions:microwave power, 2mW; microwave frequency, 9.222 GHz; microwave modulation, 2mT; temperature, 15 K. Thegain was the same for all spectra.

and C of Fig. 1 are similar, as are those of spectra Band C of Fig. 2 (see also the quantification below).In contrast, however, the spectra of samples frozen5s after mixing are significantly different in theabsence and in the presence of cyanide (comparespectra D and E of each Figure). The g = 3 signal,due to the production of ferricytochrome c byaerobic oxidation catalysed by the enzyme, is muchsmaller in the presence of cyanide. It is apparenttherefore that, during the first 5s after mixing,cyanide inhibition is fully established. Indeed,examinaton of the e.p.r. spectra (not shown) ofsamples captured I s after mixing shows that in thepresence of cyanide the g = 3 signal due to ferri-cytochrome c is developed to its maximum extentand does not increase thereafter. Thus the influ-

ence ofcyanide is developed within the first secondof turnover.A quantitative analysis of these spectra may be

obtained by integrating the signals as described inthe Materials and methods section. The results ofthis quantification are shown in Table 1 and arenow discussed below.

Resting enzyme

For the resting enzyme and within the dead timeof the apparatus approx. 1.4 electron equivalents offerricytochrome c are produced both in the absenceand in the presence of 10mM-cyanide. This is to beexpected, as it is known that a rapid burst phase ofthe reaction between ferrocytochrome c and cyto-chrome oxidase transfers 1-2 electron equivalents

1984

11111 I I I I 1' I

t~~0t- -x5

B

C

D

E

832

Cyanide inhibition of cytochrome c oxidase

12 8 6 5 4 3l-

g-value2 1.5

11111 I I

A('7B

C

D

E

x5

0.1

-/s-ViN~~~~~

0.2 0.3Magnetic flux density (T)

0.4 0.5

Fig. 2. E.p.r. spectrafrom rapid-freeze experiments with the pulsed cytochrome oxidase versus reduced cytochrome c plus 02Spectrum A: pulsed control sample, reduced cytochrome oxidase mixed with 02-bubbled Hepes buffer and frozenwithout delay; cytochrome oxidase concentration, 173pM after mixing. Spectrum B: reduced cytochrome oxidasemixed with cytochrome c2+ and 02 ; no delay after mixing; cytochrome oxidase concentration, 173 pM; cytochrome cconcentration, 1.25mm. Spectrum C: as for spectrum B but with lOmM-KCN (added in the cytochrome c syringe).Spectrum D: as for spectrum B but with 5s delay after mixing. Spectrum E: as for spectrum C but with 5 s delay aftermixing. Spectrum F: transferrin, 100pM (internal standard); the signal shown was digitally subtracted from spectraA-E. Conditions were as described for Fig. 1. Freeze-dried transferrin was added to the cytochrome oxidase-containing syringe just before the start of the experiment.

Table 1. Electron balance sheet for the reaction between ferrocytochrome c and cytochrome c oxidaseThe cytochrome C3+ values have been corrected for remaining cytochrome a3+, i.e. cytochrome C3+ = total g = 3signal -oxidized cytochrome a. * indicates that the electron distribution was unbalanced owing to transfer to 02.- indicates that the signal was too small to integrate.

Electron equivalents

Delay(s)

Resting No5

Pulsed No5

No cyanide, ~ ~ ~~AI

TotalCUB+ (e- in

c3+ a2+ CUA+ (g= 6) oxidase)1.4 > 0.9 0.75 0.2 1.854.5 0.7 0.7 *

1.854.8

0.60.7

0.45

0.450.4

0.050.13

1.1

C3+1.42.40.652.05

a2+>0.9>0.90.60.80

1.30.3 1.75

Vol. 224

With cyanide

CUB+0.15

CUA+0.70.750.70.65

a33+-HCN(g= 3.55)

0.050.2

Total(e- in

oxidase)1.81.85

833

P. Jensen, M. T. Wilson, R. Aasa and B. G. Malmstrom

from the donor in the first few milliseconds of thereaction and that this burst is insensitive to thepresence of cyanide (Wilson et al., 1975). Theelectrons from cytochrome c are distributed amongthe sites in the oxidase as indicated in Table 1. Ineach case cytochrome a is almost completelyreduced and CUA is approximately 70% reduced. Asmall axial g = 6 signal arises within this time, thispresumably being due to reduction of a portion ofthe CUB sites and thus the appearance ofuncoupledferricytochrome a3. In the presence of cyanide apart of this latter signal is converted into g = 3.55,a low-spin ferric haem-cyanide signal. The totalnumber of electron equivalents within the oxidaseis approx. 1.8, somewhat higher than the expectedvalue of 1.4 but probably equal to this value withinthe error of all the integrations involved.

After a 5s delay before quenching, the intensi-ties of the e.p.r. signals are now quite different inthe absence and in the presence of cyanide. In theformer case the enzyme has clearly 'turned over'and catalysed the oxidation of cytochrome c by 02.Thus the ferricytochrome c concentration isapprox. 4.5 electron equivalents and approachesthe maximum expected for this experimentalregime. Under these conditions (1 mM-cytochromec and 123 /M-oxidase) this maximum should beachieved when the 02 is exhausted; assumingapproximately 200 kM-02, we should see at maxi-mum approx. 6 haem equivalents of cytochrome coxidized (i.e. 200 x 4 123), and less if the 02 con-centration is lower than 200pm in the strongprotein solutions used before mixing. The electronscannot be accounted for by reduction of the siteswithin the enzyme and have presumably beenpassed on to 02. In the presence of cyanide, how-ever, only 2.4 equivalents of ferricytochrome c areseen: 1.9 of these are accounted for by the enzymesignals, whereas the remainder presumably havebeen passed to 02 before complete inhibitionensued. The g = 6 signal has now vanished, to beentirely replaced by the cyanide-bound signal atg = 3.55, which is equivalent to 0.2 haem equi-valent. This latter value is particularly interesting asit indicates that, although the enzyme is apparentlyfully inhibited, only 20% of cytochrome a3 is in thelow-spin ferric haem-cyanide complex and thus itcannot be this species that is of primary impor-tance for the inhibition. Whatever the inhibitedform is, it is seemingly e.p.r.-silent. A goodcandidate for this form is thus one in which bothCUB and cytochrome a3 remain oxidized andcoupled, possibly by cyanide acting as bridgingligand, as suggested by Hill et al. (1983).Our results on the resting enzyme are thus

consistent with the following interpretations. Onmixing the oxidized enzyme with ferrocytochromec in the presence ofcyanide (which does not bind to

the oxidized enzyme on this time scale) approx. 2electron equivalents enter 80% of the enzymemolecules. On achieving this state of partialreduction a transition occurs within the enzymethat allows cyanide to approach and bind to thebinuclear cytochrome a3-CuB site. In so doing,cyanide strongly stabilizes the oxidized forms ofthe metals to which it is tightly bound, thusstopping reduction of these sites and hencecatalysis. In some 20% of the molecules, however,electrons may transfer also on to CUB, and theuncoupled ferricytochrome a3 now becomes e.p.r.-visible and binds cyanide, again leading toinhibition. This interpretation thus supports theearlier kinetic studies which suggest that it ispartially-reduced turnover intermediates thatrapidly bind cyanide and also highlights theknown heterogeneity of the resting enzyme.

Pulsed enzymeThe pulsed enzyme behaved in an essentially

similar manner to the resting enzyme. In theabsence of cyanide approx. 2 electron equivalentsare delivered to the enzyme in the burst phase, asseen by the appearance of the ferricytochrome ce.p.r. signal. Half of these electron equivalents canbe seen in the enzyme, the remainder we assumehaving passed into 02. This situation is slightlydifferent from that of the resting enzyme, probablyreflecting the enhanced activity of the enzyme thatthe pulsed form is known to possess (Antonini etal., 1977). After 5s the enzyme has catalysed theoxidation of the remaining cytochrome c.

In the presence of cyanide the situation is a littlemore complex. After 5s delay the pattern seen forthe resting enzyme is repeated. Two electronequivalents have passed into the enzyme, which isnow fully inhibited by cyanide, although only afraction of the enzyme displays the g = 3.55 signal.With no delay, however, the picture is somewhatdifferent. A significantly smaller amount of ferri-cytochrome c is formed (approx. 0.7 electronequivalent) while almost twice the electron equi-valents are observed in the oxidase. We rationalizethis situation as follows. Immediately after themixing the enzyme for the pulsed experiments isstill fully reduced in the presence of 02 02 willnow bind and accept 2 electron equivalents witha rate in the order of 3 x 104s-. The sites in theenzyme donating these have until recently beenconsidered to be cytochrome a3 and CUB; however,Hill & Greenwood (1984) have shown that in aproportion of the molecules (40%) cytochromes aand a3 may be the donors. In either case theenzyme will now exist with cytochrome a3 andpossibly CUB oxidized. Further electron transferfrom CUA and cytochrome a to the 02-binding siteis slower, with a rate of approx. 5 x 103s- , and

1984

834

Cyanide inhibition of cytochrome c oxidase

Table 2. Electron equivalents remaining in cytochrome aand CUA

Reduced cytochrome oxidase (246pM) was mixedwith aerated 5OmM-Hepes buffer, pH7.4, at roomtemperature, and freeze-quenched in cold isopen-tane. The sample without cyanide (OmM) was usedas the reference and the e.p.r. signals from thecompletely re-oxidized cytochrome a and CUA inthis sample were measured (see the Materials andmethods section). The e.p.r. signals obtained aftermixing the cytochrome oxidase solution with cyan-ide-containing aerated buffer (10mM- and 30mM-cyanide) were smaller than in the reference, and thisdifference was taken as a measure of the number ofelectron equivalents remaining in those samples.

Concn. ofCN- (mM)

0

1030

Electron equivalents

a2+ CUA+o o

0.2 0.150.4 0.4

which is therefore comparable with the rate ofcyanide binding. At 10mM-cyanide and taking arate constant of approx. 106 M-1 s1 for the second-order combination rate constant of cyanide to thepartially reduced enzyme (Jones et al., 1984), thecyanide 'on' rate is approx. 103-104 s-1. Thus theremay be competition between cyanide binding toand reduction of the oxidized sites in cytochromeoxidase. If this is so, then we may expect that in afraction of the molecules cyanide will bind to theenzyme and stop electron transfer from cyto-chrome a and CUA. The electrons that we observeon cytochrome a and CUA (Table 1, row 3: pulsed,no delay, plus cyanide) are therefore only partiallyderived from cytochrome c and are in fact electronsthat have remained in the enzyme from theoriginal reduction before the mixing. This alsoexplains why we see more electron equivalents inthe enzyme than we see cytochrome c oxidized.

Support for this explanation comes from controlrapid-freeze experiments in which the fully re-duced enzyme was mixed with oxygenated buffereither in the absence or in the presence of cyanide.Table 2 shows that in the absence of cyanide all thesites are oxidized within the dead time of theapparatus. In the presence of cyanide, however,the oxidation of cytochrome a and CUA is partiallyblocked, and this effect is exaggerated at highercyanide concentrations. Such observations areconsistent with the kinetic competition for sites asdiscussed above.

Reduction ofcytochrome c oxidase by dithionite in thepresence and in the absence of cyanide

Parallel experiments were carried out monitor-ing the reduction of cytochrome oxidase by

g-value12 8 6 5 4 3 2

0.1 0.2 0.3Magnetic flux density (T)

Fig. 3. E.p.r. spectra of cytochrome oxidase after freeze-thaw experiments with cyanide present

Spectrum A: cytochrome oxidase, 220/AM, frozenlOs after the addition of Na2S204 (6.5mM) byimmersing the sample in cold (140K) isopentane('fast freeze'). Spectrum B: as for spectrum A butwith KCN (0.5 mM) added together with thedithionite. Spectrum C: the sample used for spec-trum B (above) thawed to room temperature for1 min and re-frozen ('fast freeze'). Spectrum D: thesample used for spectrum C after an additional10min thawing. Conditions were as described forFig. 1.

dithionite either by optical or e.p.r. spectroscopicmethods.At a wavelength of 445nm and on addition of

dithionite (25 mM) a rapid reduction of a portion ofthe enzyme (cytochrome a) was followed by slowreduction ofcytochrome a3 (Jones et al., 1983) witha t4 of approx. 150s. In the presence of 300pM-cyanide the rapid phase of reduction remainedunchanged, but the rate of the slower phase was

greatly lowered, ti approx. 1500s. These results arein agreement with those reported by Jones et al.(1984) and show that cyanide binds rapidly to thepartially reduced enzyme, stopping its full reduc-tion. The rate of reduction of cytochrome a3 is nowlimited by the cyanide 'of rate.

Fig. 3 reports the complementary e.p.r. spectra.

Vol. 224

1111__ I II ,I

A

D-v110~~~~~~~

835

P. Jensen, M. T. Wilson, R. Aasa and B. G. Malmstrom

Spectrum A is that obtained from an anaerobic(N2-equilibrated) sample of resting oxidase intowhich 6.5mM-Na2S204 was mixed and the samefrozen after 10s. By this time the optical spectrumindicates that cytochrome a is largely reducedwhereas cytochrome a3 is oxidized. The e.p.r.spectrum (A) is in agreement with the opticalmeasurements, the g = 3 signal (cytochrome a) isequivalent to only approx. 0.1 haem equivalentand the CUA is similarly small. The g= 6 signal(ferric cytochrome a3) comprises about 0.2 haemequivalent and the remainder of the cytochrome a3is presumably still oxidized and coupled to CUB.SpectrumB shows the same experiment carried outin the presence of 0.5 mM-KCN. Although optical(see above) and activity (Jones et al., 1984)measurements suggest that the enzyme is now fullycomplexed with cyanide, we observe a g = 3.55signal (cytochrome a33+-HCN) comprising onlyabout 0.15 haem equivalent, and the g = 6 signalhas decreased by the same amount. On waiting fora further period (spectrum C), no additional signalsappear, but the remaining g = 6 signal is convertedinto g = 3.55. The elapse of further time leads to nochanges except that the character of the g = 3.55signal is altered and the g = 12 signal disappears.Although we are at present uncertain of thesignificance of these changes, the observations arein agreement with those made by Johnson et al.(1981) and Brudvig et al. (1981), and we do notbelieve they alter our overall conclusions. Theresults of the optical and e.p.r. studies on dithionitereduction are therefore in full agreement with therapid-freeze investigation with the native reduc-tant, cytochrome c, namely that cyanide bindingand inhibition gives an e.p.r.-silent species poss-ibly involving binding to coupled ferricytochromea3 and cupric CUB. The heterogeneity observed inthe rapid-freeze experiments is again evident inFig. 3, with about 20% of the enzyme rapidlyaccepting electrons into CUB, rendering cyto-chrome a3 e.p.r.-visible. This latter then bindscyanide to yield the g = 3.55 signal.

Concluding remarks

The behaviour of cytochrome a3 (classicallydefined as the binding site of the oxidase) towardsligands is clearly strongly dependent not only on itsvalence state but also on the valence state of othermetal centres and on the conformational status ofthe enzyme. When the fully reduced enzyme reactswith ligands, e.g. CO, CN, it does so in reactionsthat are second-order and whose equilibriumassociation constants measured by static titrationsare in accord with those calculated from determin-ations of the 'on' and 'off' rates. In other wordsthe ligand-binding site is accessible as it is in open

and rapid communication with the bulk-phasesolution. The same is true also for 02, which bindsexceedingly rapidly (k.. 108 M-1 * s- I) to the cyto-chrome a3-CuB binuclear centre.

Ligand binding to the fully oxidized enzyme is,in contrast, complex. For example, althoughcyanide binds very tightly to cytochrome a3(KD 10-6 M), it does so very slowly and with a rateindependent of cyanide concentration and witha high temperature coefficient. Such behavioursuggests that the binding site is shielded fromapproaching ligands and that complex-formationis rate-limited by transitions in the protein,possibly conformational, as suggested by the highenthalpy of activation (Van Buuren et al., 1972),which allow ligands from the bulk phase toapproach the binding site.The transition from a 'closed' to an 'open' site is

particularly important to the consideration ofcyanide inhibition. It has been shown (Jones et al.,1984; Van Buuren, 1972) that it is binding ofcyanide to partially reduced forms of the enzyme,populated during turnover, that is responsible forthe powerful inhibiting effects of cyanide. In theseforms of the enzyme it appears that cytochrome a3remains oxidized yet is accessible to cyanide,as indicated by the rapid onset of inhibition(Antonini et al., 1971). In a recent study (Jones etal., 1984) a model was proposed that incorporatedthis transition from a 'closed' to an 'open' form onpartial reduction. However, it was impossible toascertain from the kinetic experiments the degreeof reduction necessary to initiate this transition.The present study suggests that the entry of 2electron equivalents into the enzyme from cyto-chrome c is sufficient to switch the cytochrome a3site to the accessible form. These electrons resideon cytochrome a and CUA, and thus it appears thatreduction of the electron-accepting site is commu-nicated to the ligand-binding site, which is nowable to bind cyanide rapidly. The e.p.r. measure-ments also show that in the cyanide-bound form(apart from in a small fraction of the enzyme)cytochrome a33+ is still coupled to CUB2+, and it islikely that the ligand is stabilizing both metals intheir oxidized states by bridging between them, assuggested by Hill et al. (1983).

Although entry of 2 electron equivalents issufficient to trigger the 'opening' of the ligand-binding site, we cannot be certain that both arenecessary. The time-resolution of our method doesnot allow us to discount the possibility thatreduction of one of the sites, cytochrome a or CuA,is sufficient.

In contrast, a common view has been that aninput of 3 electron equivalents is necessary forcyanide binding (see, e.g., Nicholls, 1983). Thisdivergence can, however, be explained by consid-

1984

836

Cyanide inhibition of cytochrome c oxidase 837

ering the different proportions ofenzyme forms, byusing the criteria of Brudvig et al. (1981), inoxidase prepared by the methods of Yonetani andVan Buuren (Wilson et al., 1982) respectively.These differences are also reflected in the size ofthe g = 3.55 e.p.r. signal from the a33+-HCNcompex, which amounts to the equivalent of 1haem with a Yonetani preparation (Johnson et al.,1981) whereas with an enzyme preparation bythe method of Van Buuren only about 20% ofthe molecules are in the form giving a g = 3.55signal.The nature of the transition leading to an open

configuration may involve gross conformationalchanges [as suggested by the occluded nature of theresting site (see above) and by the change in thepacking characteristics in the rapid-freeze experi-ments (see the Materials and methods section)], adisplacement of an intrinsic ligand from the sixthco-ordination site of the haem ofcytochrome a3, ormost probably both of these changes. For examplefully oxidized pulsed enzyme, though being muchmore accessible to cyanide than the resting form(Brittain & Greenwood, 1976; Jones et al., 1984), iscertainly not as open to attack by cyanide as arepartially reduced turnover forms.The 'opening' and 'closing' of the site to ligands

presumably has a functional significance for theenzyme in vivo. An obvious possibility is that itconstitutes a device whereby 02 may only ap-proach the binding site when two metal centres arereduced. Thus, once electrons have passed to bothCUB and cytochrome a3, the °2 may bind and thefirst 2-electron reduction to form tightly boundperoxide may proceed. A second possibility is thatsuch a conformational transition constitutes partofthe energy transduction system, e.g. the mechan-ical part of a proton pump.

This investigation was supported by the SwedishNatural Science Research Council. A British CouncilTravel Grant is also acknowledged (M. T. W.). M. T. W.also acknowledges Science and Engineering ResearchCouncil Support. We thank Professor Tore VanngArd forvaluable suggestions and discussions, and Mr. Per-EricThornstr6m, Mrs. Kajsu Valimaa and Mrs. Elsy Lager-dahl for skilful technical assistance during the enzymepreparations and the ensuing characterizations.

ReferencesAasa, R. & Vainngard, T. (1975) J. Magn. Reson. 19, 308-

315

Aasa, R., Malmstrom, B. G., Saltman, P. & VanngArd,T. (1963) Biochim. Biophys. Acta 75, 203-222

Aasa, R., Albracht, S. P. J., Falk, K.-E., Lanne, B. &Vanngard, T. (1976) Biochim. Biophys. Acta 422, 260-272

Antonini, E., Brunori, M., Greenwood, C., Malmstrom,B. G. & Rotilio, G. C. (1971) Eur. J. Biochem. 23, 396-400

Antonini, E., Brunori, M., Colosimo, A., Greenwood, C.& Wilson, M. T. (1977) Proc. Natl. Acad. Sci. U.S.A.74, 3128-3132

Ballou, D. P. (1971) Ph.D. Thesis, University ofMichigan

Ballou, D. P. & Palmer, G. (1974) Anal. Chem. 46, 1248-1253

Brittain, T. & Greenwood, C. (1976) Biochem. J. 155,453-455

Brudvig, G. W., Stevens, T. H., Morse, R. H. & Chan,S. I. (1981) Biochemistry 20, 3912-3921

Brunori, M., Antonini, E. & Wilson, M. T. (1981) inMetal Ions in Biological Systems (Segel, H., ed.), vol.13, pp. 187-228, Marcel Dekker, New York

De Vries, S. & Albracht, S. P. J. (1979) Biochim. Biophys.Acta 546, 334-340

Downer, N. W., Robinson, N. C. & Capaldi, R. A.(1976) Biochemistry 15, 2930-2936

Hill, B. & Greenwood, C. (1984) FEBS Lett. 166, 362-366

Hill, B., Brittain, T., Eglinton, D. G., Gadsby, P. M. A.,Greenwood, C., Nicholls, P., Peterson, J., Thomson,A. J. & Woon, T. C. (1983) Biochem. J. 215, 57-66

Jensen, P., Aasa, R. & Malmstr6m, B. G. (1981) FEBSLett. 125, 161-164

Johnson, M. K., Eglinton, D. G., Gooding, P. E.,Greenwood, C. & Thomson, A. J. (1981) Biochem. J.193, 699-708

Jones, G. D., Jones, M. G., Wilson, M. T., Brunori, M.,Colosimo, A. & Sarti, P. (1983) Biochem. J. 209, 175-182

Jones, M. G., Bickar, D., Wilson, M. T., Brunori, M.,Colosimo, A. & Sarti, P. (1984) Biochem. J. 220, 57-66

Nicholls, P. (1983) Trends Biochem. Sci. 8, 353-354Sekuzu, I., Takemori, S., Yonetani, T. & Okunuki, K.

(1959) J. Biochem. (Tokyo) 46, 43Smith, L. & Conrad, H. (1956) Arch. Biochem. Biophys.

63, 403-413Van Buuren, K. J. H. (1972) Ph.D. Thesis, University ofAmsterdam

Van Buuren, K. J. H., Nicholls, P. & Van Gelder, B.(1972) Biochim. Biophys. Acta 256, 258-276

Wikstrom, M., Krab, K. & Saraste, M. (1981) Cyto-chrome Oxidase: A Synthesis, pp. 14-54, AcademicPress, New York

Wilson, M. T., Greenwood, C., Brunori, M. & Antonini,E. (1975) Biochem. J. 147, 145-153

Wilson, M. T., Jensen, P., Aasa, R., Malmstr6m, B. G. &Vanngard, T. (1982) Biochem. J. 203, 483-492

Vol. 224